Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

LLM2Rec: Large Language Models Are Powerful Embedding Models for Sequential Recommendation

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read LLM2Rec claims that fine-tuning a large language model on user interaction histories yields item embeddings that combine semantics with collaborative-filtering signals and outperform text-embedding baselines in sequential recommendation.

desk verdict A solid and reproducible empirical pipeline for injecting collaborative signals into LLM embeddings, but the mechanism is asserted rather than proven and the evaluation dodges ID-based baselines. read the letter →

arxiv 2506.21579 v1 pith:FONGVI2V submitted 2025-06-16 cs.IR cs.AI

Yingzhi He , Xiaohao Liu , An Zhang , Yunshan Ma , Tat-Seng Chua This is my paper · ORCID
classification cs.IRcs.AI
keywords SequentialrecommendationLargelanguagemodelsItemembeddingsCollaborativefilteringContrastivelearningSupervisedfine-tuningOut-of-domaingeneralizationBidirectionalattention
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that an ideal item embedding for sequential recommendation must carry two kinds of information at once: what an item is (semantics) and how it co-occurs with other items in user histories (collaborative filtering, or CF). It proposes LLM2Rec, a two-stage training recipe that first fine-tunes an LLM to predict the next item in real user sequences, then reshapes it into an embedding model with bidirectional attention, masked token prediction, and item-level contrastive learning. The reported experiments show that embeddings from this recipe improve downstream recommenders over all tested text-embedding baselines, on datasets from the training categories and on categories and platforms the model never saw. That matters because conventional recommenders either capture CF but fail to transfer to new domains, or capture semantics but ignore CF; LLM2Rec is an attempt to get both in a single embedding space.

What carries the argument

The load-bearing object is the trained item embedder $E = \mathrm{avg} \circ \pi_\theta$: an LLM whose causal attention has been replaced by bidirectional attention, whose output is mean-pooled over title tokens, and whose weights are obtained by two sequential training stages. CSFT is the stage that injects CF signals: the model is supervised to autoregressively generate the next item title in a user's sequence, which forces hidden states of item titles to reflect co-occurrence statistics. IEM then reconciles this CF-aware model with embedding-style inference: MNTP adapts the weights to bidirectional attention, and item-level contrastive learning with randomly masked views pulls augmentations of the same item together while separating distinct items, producing distinguishable item-level vectors.

What would settle it

Replace each item title in the fine-tuning sequences with a random but consistent token, or shuffle the item order within each sequence, while keeping everything else fixed; if the resulting embeddings still produce the reported gains on the out-of-domain datasets, then co-occurrence structure is not what carries the improvement. A direct geometry check—measuring whether co-purchased item pairs become closer in LLM2Rec's embedding space than in the base LLM's space—would make the CF claim observable.

Watch

Extended reading notes

Core claim

The central claim is that a decoder-only LLM can be turned into a recommendation embedding model that is simultaneously semantic and collaborative. The mechanism is the order of training: first, Collaborative Supervised Fine-Tuning (CSFT) makes the LLM predict the next item given a user's previous item titles, exposing the hidden states to CF-driven co-occurrence patterns; second, Item-level Embedding Modeling (IEM) replaces causal attention with bidirectional attention, adds masked next-token prediction, and applies item-level contrastive learning so the model produces fixed-sized item embeddings rather than token predictions. The paper reports that LLM2Rec consistently outperforms all compared baselines on both in-domain and out-of-domain datasets, with the largest gains on out-of-domain categories, and that the CSFT stage contributes the largest share of the improvement in the ablation.

Load-bearing premise

The load-bearing premise is that making the LLM predict the next item in real user sequences really does bake 'items that tend to be bought together' into the hidden states of individual item titles, and that this information survives the later switch to bidirectional attention and averaging.

Editorial extensions

If this is right

  • Sequential recommenders can be improved without learning item IDs from scratch: a fixed, general embedding from LLM2Rec, plus a small linear adapter, outperforms both pure text embeddings and recommendation-specific embedding models.
  • CF signals can be transferred across domains: because the embedding is trained on mixed categories, items from categories unseen at training time still receive embeddings that carry useful behavioral information.
  • The two-stage order matters: causal attention fine-tuning for recommendation, followed by a bidirectional embedding adaptation, is what yields the gains; reversing or skipping the collaborative stage degrades performance.
  • Small LLM backbones suffice for practical gains: the reported results are based on a 0.5B-parameter backbone, with larger backbones improving further but at higher cost.
  • Dataset diversity in pre-training improves out-of-domain generalization, while a focused single-category pre-training can be better for that same category.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if CSFT truly embeds CF structure, then LLM2Rec embeddings should reconstruct known co-purchase or co-click graphs better than text embeddings; a direct nearest-neighbor test on held-out interaction pairs would make the claim's mechanism visible.
  • Editorial inference: the framework suggests a path to a single 'universal' item-embedding model trained once on many platforms and categories, then frozen for downstream recommenders; the paper's cross-platform result is a first step but larger cross-platform evaluations would test it.
  • Editorial inference: the same two-stage recipe—supervised collaborative fine-tuning followed by bidirectional embedding adaptation—could generalize to other structured prediction tasks where an embedding must encode both content and relational context, such as citation recommendation or link prediction on text-attributed networks.
  • Editorial inference: because the reported gains rest on an ablation rather than on controlled perturbation of sequence statistics, a natural stress test is to shuffle item order or replace item titles with random identifiers during CSFT; if gains persist, the credit may be vocabulary or domain adaptation rather than CF structure.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. This paper introduces LLM2Rec, a two-stage framework for turning a decoder-only LLM into an item embedding model for sequential recommendation. In Stage 1 (Collaborative Supervised Fine-Tuning, CSFT), the LLM is fully fine-tuned for 10,000 steps on 1.03M user interactions from six Amazon categories to predict the next item title given the titles of previous interactions. In Stage 2 (Item-level Embedding Modeling, IEM), the model is converted to bidirectional attention, trained with masked next-token prediction and item-level SimCSE-style contrastive learning, and item embeddings are obtained by mean pooling. These embeddings are passed through a linear adapter into GRU4Rec and SASRec. Experiments cover three in-domain Amazon datasets (Games, Arts, Movies) and three out-of-domain datasets (Sports, Baby, Goodreads), with Recall@10/20 and NDCG@10/20. The authors report consistent gains over BERT, GTE, BGE, LLM2Vec, BLAIR, EasyRec, and LLMEmb, and provide ablations, backbone studies, mixed-dataset studies, and efficiency analysis.

Significance. If the results are taken at face value, the paper makes a useful practical contribution: a lightweight recipe (Qwen2-0.5B backbone) for producing recommendation-aware embeddings that improve downstream sequential recommenders over existing text embeddings, including on categories never seen during pretraining. The evidence for the narrow claim is solid: Table 3 reports a fair text-embedding comparison under a fixed protocol with three-seed averaging, and Table 5 and Figure 4 support the generalization and backbone claims. The code is promised and the experimental setup follows standard datasets and full-ranking evaluation. However, the broader framing rests on two unsupported pillars: the absence of ID-based recommenders as anchors, and the claim that CSFT specifically encodes collaborative-filtering structure that survives mean-pooled item embeddings. Both are testable and should be addressed before the paper can support its title-level claim.

major comments (3)
  1. [§4.1.2, Table 3] The evaluation compares only text-embedding-initialized recommenders; no ID-based sequential recommender trained from scratch is reported. Because ID-based SASRec and GRU4Rec are the dominant paradigm for sequential recommendation, the absence of these numbers means Table 3 cannot establish the paper's title-level claim that LLM2Rec is a 'powerful embedding model for sequential recommendation' in an absolute sense. It only establishes superiority over the listed text-embedding baselines. Please add ID-based GRU4Rec and SASRec on the same data splits under the same full-ranking protocol, or explicitly reframe the contribution as 'text-embedding models for sequential recommendation' and soften the title and abstract accordingly.
  2. [§3.2–§3.3, Eq. (2)–(4), Table 4] The central mechanism — that CSFT injects item-level collaborative-filtering (CF) signals and that these signals survive IEM — is not directly evidenced. The only quantitative support is the downstream accuracy ablation in Table 4, which is a confounded proxy: fine-tuning on next-item-title prediction over 1.03M interactions can improve accuracy by adapting to item-title vocabulary, category-transition regularities, or popularity, without inducing pairwise co-occurrence geometry in single-item embeddings. There is no control that destroys interaction co-occurrence while preserving token distributions (e.g., shuffling items within sequences or permuting user-item assignments), and Eq. (3)–(4) contain no sequence-level objective, so the persistence of CF information is assumed. Moreover, Sports, Baby, and Goodreads items have no interaction history in the pretraining mixture, so CSFT gains on these datasets (e.g., Sports R@10 from 0.1079 to 0.1119 in Table 4) cannot stem from item-level CF; they are more plausibly explained by transferable category or semantic regularities. Please add a direct test of embedding geometry (e.g., co-purchased pairs closer than category-matched non-co-purchased pairs), a shuffled-co-occurrence control, and a discussion of how CSFT could benefit out-of-domain items.
  3. [Table 4, rows 'Causal' and 'Bidirectional'] The comparison between the 'Causal' and 'Bidirectional' settings changes both the attention mask and the pooling strategy: the causal row uses the last hidden state of the [EOS] token, while the bidirectional row uses mean pooling over all token hidden states. The reported improvement therefore cannot be attributed to bidirectional attention alone, which is the stated design motivation for IEM. A controlled comparison should hold the pooling strategy fixed, e.g., causal attention with mean pooling versus bidirectional attention with mean pooling, to isolate the effect of the attention mask.
minor comments (6)
  1. [Table 4] The header 'Casual' should be 'Causal'.
  2. [§3.2] The sentence 'we only retrain the item titles with some necessary separators, like commas' appears to be a typo; 'retrain' should likely be 'retain' or 'keep', since the surrounding text describes restricting the input to item titles rather than training the titles.
  3. [§4.4.1] The text 'LLM2Rec consistently outperforms the general-purpose embedding baseline, LLM2Rec' should refer to 'LLM2Vec' rather than 'LLM2Rec'.
  4. [References] References [23] and [24] are the same paper; please deduplicate the bibliography and adjust the in-text citations accordingly.
  5. [§4.1.1] The heading 'Evaluation Mertrics' contains a typo and should read 'Evaluation Metrics'.
  6. [Algorithm 1] Algorithm 1 uses epoch counts E1, E2, and E3, but Section 4.1.3 reports step counts (10,000, 1,000, 1,000); please align the two descriptions for reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the evaluation uses held-out interactions, hyperparameters follow external prior work, and the CF claim is an empirical interpretation rather than a definitional reduction.

full rationale

LLM2Rec is an empirical method paper rather than a derivation from first principles, so there is no derivation chain whose output is equivalent to its input by construction. The central training objective L_CSFT (Eq. 2) predicts next-item titles from historical interaction sequences on the training split; all reported results are on held-out validation/test interactions created by leave-one-out partitioning, so the reported numbers are not the fitted values. The IEM objectives (Eqs. 3 and 4) are masked next-token prediction and SimCSE-style contrastive learning, with temperature, dropout, and mask ratio taken from external prior work (LLM2Vec, SimCSE) rather than calibrated to the target datasets. The claim that CSFT injects collaborative-filtering signals is an empirical interpretation supported by an illustration and an ablation; even though the ablation does not fully isolate CF structure from vocabulary adaptation, that is a validity limitation, not circularity. Self-citations in the related work (e.g., refs. 6, 31, 41, 45, 53, 54) are not load-bearing: the claims they support are also backed by external references and are not uniqueness theorems or forbidden alternatives. The in-domain setup pre-trains on the same training splits later used by the downstream recommender, which is a data-comparability concern, but the out-of-domain Sports/Baby/Goodreads results provide independent support for generalization, and no parameter is fitted to those results. No step reduces, by equation or by self-citation, to its own input.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters were fitted to the reported test results; the only hyperparameters are standard contrastive learning settings inherited from prior work. The paper adds no new physical or conceptual entities, only a training procedure. The axioms above are the unproven domain assumptions that the empirical evaluation rests on.

assumptions (3)
  • domain assumption Item titles are a sufficient textual representation of items for capturing both semantics and collaborative co-occurrence patterns.
    The entire pipeline (CSFT, MNTP, contrastive learning) operates only on item titles; no other item metadata is used (Section 3.1). If titles are too ambiguous or noisy, the CF signal cannot be grounded.
  • domain assumption The order of user interactions carries the collaborative signal, and next-item prediction on ordered title sequences induces CF-aware item representations.
    This is the core mechanism claimed in Section 3.2; the paper provides an illustrative example (Figure 1) and an ablation (Section 4.3), but no direct measurement of what CSFT changes in the embedding geometry.
  • domain assumption A linear adapter plus a standard sequential recommender is a faithful way to evaluate the quality of frozen item embeddings.
    The paper uses one linear layer and trains GRU4Rec/SASRec on top (Section 3.4). The ranking quality is then attributed to the embeddings, but a stronger adapter or a different recommender might change the ranking of methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLM2Rec: Large Language Models Are Powerful Embedding Models for Sequential Recommendation." pith.science (2026). https://pith.science/paper/FONGVI2V

@misc{pith2026250621579,
  author       = {Pith},
  title        = {Pith review of: LLM2Rec: Large Language Models Are Powerful Embedding Models for Sequential Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FONGVI2V}},
  note         = {Machine review of arXiv:2506.21579}
}
read the original abstract

Sequential recommendation aims to predict users' future interactions by modeling collaborative filtering (CF) signals from historical behaviors of similar users or items. Traditional sequential recommenders predominantly rely on ID-based embeddings, which capture CF signals through high-order co-occurrence patterns. However, these embeddings depend solely on past interactions, lacking transferable knowledge to generalize to unseen domains. Recent advances in large language models (LLMs) have motivated text-based recommendation approaches that derive item representations from textual descriptions. While these methods enhance generalization, they fail to encode CF signals-i.e., latent item correlations and preference patterns-crucial for effective recommendation. We argue that an ideal embedding model should seamlessly integrate CF signals with rich semantic representations to improve both in-domain and out-of-domain recommendation performance. To this end, we propose LLM2Rec, a novel embedding model tailored for sequential recommendation, integrating the rich semantic understanding of LLMs with CF awareness. Our approach follows a two-stage training framework: (1) Collaborative Supervised Fine-tuning, which adapts LLMs to infer item relationships based on historical interactions, and (2) Item-level Embedding Modeling, which refines these specialized LLMs into structured item embedding models that encode both semantic and collaborative information. Extensive experiments on real-world datasets demonstrate that LLM2Rec effectively improves recommendation quality across both in-domain and out-of-domain settings. Our findings highlight the potential of leveraging LLMs to build more robust, generalizable embedding models for sequential recommendation. Our codes are available at https://github.com/HappyPointer/LLM2Rec.

Figures

Figures reproduced from arXiv: 2506.21579 by the authors.

Figure 1
Figure 1. An illustrating example that pre-trained language [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the overall pre-training framework of LLM2Rec and how the generated embeddings are utilized for [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. An example of the collaborative instruction for [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Performance comparison of embedding methods across different LLM backbones. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Comparison of inference time and performance of [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Beyond Semantic Understanding: Preserving Collaborative Frequency Components in LLM-based Recommendation

    cs.CL 2025-08 conditional novelty 6.0 of 10

    FreLLM4Rec shows that LLMs attenuate low-frequency collaborative components of item embeddings and introduces graph and temporal low-pass filters that preserve them, achieving up to 8% NDCG@10 gains.

Reference graph

Works this paper leans on

70 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    Keqin Bao, Jizhi Zhang, Wenjie Wang, Yang Zhang, Zhengyi Yang, Yancheng Luo, Chong Chen, Fuli Feng, and Qi Tian. 2023. A bi-step grounding paradigm for large language models in recommendation systems.arXiv preprint arXiv:2308.08434 (2023)

    arXiv 2023

  2. [2]

    Keqin Bao, Jizhi Zhang, Yang Zhang, Wenjie Wang, Fuli Feng, and Xiangnan He

  3. [3]

    Parishad BehnamGhader, Vaibhav Adlakha, Marius Mosbach, Dzmitry Bahdanau, Nicolas Chapados, and Siva Reddy. 2024. Llm2vec: Large language models are secretly powerful text encoders.arXiv preprint arXiv:2404.05961(2024)

    arXiv 2024

  4. [4]

    Gal Chechik, Varun Sharma, Uri Shalit, and Samy Bengio. 2010. Large scale online learning of image similarity through ranking.Journal of Machine Learning Research11, 3 (2010)

    work page 2010

  5. [5]

    Junyi Chen, Lu Chi, Bingyue Peng, and Zehuan Yuan. 2024. Hllm: Enhancing sequential recommendations via hierarchical large language models for item and user modeling.arXiv preprint arXiv:2409.12740(2024)

    arXiv 2024

  6. [6]

    Yuxin Chen, Junfei Tan, An Zhang, Zhengyi Yang, Leheng Sheng, Enzhi Zhang, Xiang Wang, and Tat-Seng Chua. 2024. On softmax direct preference optimization for recommendation.arXiv preprint arXiv:2406.09215(2024)

    arXiv 2024

  7. [7]

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Imagenet: A large-scale hierarchical image database. In2009 IEEE conference on computer vision and pattern recognition. Ieee, 248–255

    2009

  8. [8]

    Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. Simcse: Simple contrastive learning of sentence embeddings.arXiv preprint arXiv:2104.08821(2021)

    arXiv 2021

Show all 70 references
  1. [9]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. InSIGIR. 639–648

    2020

  2. [10]

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. 2017. Neural collaborative filtering. InWWW. 173–182

    2017

  3. [11]

    Dan Hendrycks, Steven Basart, Norman Mu, Saurav Kadavath, Frank Wang, Evan Dorundo, Rahul Desai, Tyler Zhu, Samyak Parajuli, Mike Guo, et al. 2021. The many faces of robustness: A critical analysis of out-of-distribution generalization. InProceedings of the IEEE/CVF internatio...

    2021

  4. [12]

    Balázs Hidasi and Alexandros Karatzoglou. 2018. Recurrent neural networks with top-k gains for session-based recommendations. InCIKM. 843–852

    2018

  5. [13]

    Yupeng Hou, Jiacheng Li, Zhankui He, An Yan, Xiusi Chen, and Julian McAuley

  6. [14]

    Yupeng Hou, Shanlei Mu, Wayne Xin Zhao, Yaliang Li, Bolin Ding, and Ji-Rong Wen. 2022. Towards universal sequence representation learning for recommender systems. InSIGKDD. 585–593

    2022

  7. [15]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recom- mendation. InICDM. IEEE, 197–206

    2018

  8. [16]

    Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open- domain question answering.arXiv preprint arXiv:2004.04906(2020)

    2020 arXiv

  9. [17]

    Jacob Devlin Ming-Wei Chang Kenton and Lee Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of naacL-HLT, Vol. 1. Minneapolis, Minnesota

    2019

  10. [18]

    Chankyu Lee, Rajarshi Roy, Mengyao Xu, Jonathan Raiman, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. 2024. NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models.arXiv preprint arXiv:2405.17428 (2024)

    2024 arXiv

  11. [19]

    Jinhyuk Lee, Zhuyun Dai, Xiaoqi Ren, Blair Chen, Daniel Cer, Jeremy R Cole, Kai Hui, Michael Boratko, Rajvi Kapadia, Wen Ding, et al. 2024. Gecko: Versatile text embeddings distilled from large language models.arXiv preprint arXiv:2403.20327 (2024)

    2024 arXiv

  12. [20]

    Mike Lewis. 2019. Bart: Denoising sequence-to-sequence pre-training for nat- ural language generation, translation, and comprehension.arXiv preprint arXiv:1910.13461(2019)

    2019 arXiv

  13. [21]

    Chaofan Li, MingHao Qin, Shitao Xiao, Jianlyu Chen, Kun Luo, Yingxia Shao, Defu Lian, and Zheng Liu. 2024. Making text embedders few-shot learners.arXiv preprint arXiv:2409.15700(2024)

    2024 arXiv

  14. [22]

    Jiacheng Li, Ming Wang, Jin Li, Jinmiao Fu, Xin Shen, Jingbo Shang, and Julian McAuley. 2023. Text is all you need: Learning language representations for sequential recommendation. InSIGKDD. 1258–1267

    2023

  15. [24]

    Zehan Li, Xin Zhang, Yanzhao Zhang, Dingkun Long, Pengjun Xie, and Meishan Zhang. 2023. Towards general text embeddings with multi-stage contrastive learning.arXiv preprint arXiv:2308.03281(2023)

    2023 arXiv

  16. [25]

    Jianghao Lin, Rong Shan, Chenxu Zhu, Kounianhua Du, Bo Chen, Shigang Quan, Ruiming Tang, Yong Yu, and Weinan Zhang. 2024. Rella: Retrieval-enhanced large language models for lifelong sequential behavior comprehension in recom- mendation. InWWW. 3497–3508

    2024

  17. [26]

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. InComputer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Procee...

    2014

  18. [27]

    Dugang Liu, Shenxian Xian, Xiaolin Lin, Xiaolian Zhang, Hong Zhu, Yuan Fang, Zhen Chen, and Zhong Ming. 2024. A Practice-Friendly Two-Stage LLM-Enhanced Paradigm in Sequential Recommendation.arXiv preprint arXiv:2406.00333(2024)

    2024 arXiv

  19. [28]

    Qijiong Liu, Nuo Chen, Tetsuya Sakai, and Xiao-Ming Wu. 2024. Once: Boosting content-based recommendation with both open-and closed-source large language models. InSIGKDD. 452–461

    2024

  20. [29]

    Qidong Liu, Xian Wu, Wanyu Wang, Yejing Wang, Yuanshao Zhu, Xiangyu Zhao, Feng Tian, and Yefeng Zheng. 2024. Large language model empowered embedding generator for sequential recommendation.arXiv preprint arXiv:2409.19925(2024)

    2024 arXiv

  21. [30]

    Qidong Liu, Xian Wu, Xiangyu Zhao, Yejing Wang, Zijian Zhang, Feng Tian, and Yefeng Zheng. 2024. Large Language Models Enhanced Sequential Recommenda- tion for Long-tail User and Item.arXiv preprint arXiv:2405.20646(2024)

    2024 arXiv

  22. [31]

    Xiaohao Liu, Zhulin Tao, Jiahong Shao, Lifang Yang, and Xianglin Huang. 2022. Elimrec: Eliminating single-modal bias in multimedia recommendation. InPro- ceedings of the 30th ACM International Conference on Multimedia. 687–695

    2022

  23. [32]

    Yinhan Liu. 2019. Roberta: A robustly optimized bert pretraining approach.arXiv preprint arXiv:1907.11692364 (2019)

    2019 arXiv

  24. [33]

    I Loshchilov. 2017. Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101(2017)

    2017 arXiv

  25. [34]

    Bhaskar Mitra, Fernando Diaz, and Nick Craswell. 2017. Learning to match using local and distributed representations of text for web search. InProceedings of the 26th international conference on world wide web. 1291–1299

    2017

  26. [35]

    Niklas Muennighoff. 2022. Sgpt: Gpt sentence embeddings for semantic search. arXiv preprint arXiv:2202.08904(2022)

    2022 arXiv

  27. [36]

    Niklas Muennighoff, Hongjin Su, Liang Wang, Nan Yang, Furu Wei, Tao Yu, Aman- preet Singh, and Douwe Kiela. 2024. Generative representational instruction tuning.arXiv preprint arXiv:2402.09906(2024)

    2024 arXiv

  28. [37]

    Niklas Muennighoff, Nouamane Tazi, Loïc Magne, and Nils Reimers. 2022. MTEB: Massive text embedding benchmark.arXiv preprint arXiv:2210.07316(2022)

    2022 arXiv

  29. [38]

    Jianmo Ni, Chen Qu, Jing Lu, Zhuyun Dai, Gustavo Hernández Ábrego, Ji Ma, Vincent Y Zhao, Yi Luan, Keith B Hall, Ming-Wei Chang, et al. 2021. Large dual encoders are generalizable retrievers.arXiv preprint arXiv:2112.07899(2021)

    2021 arXiv

  30. [39]

    Xubin Ren and Chao Huang. 2024. EasyRec: Simple yet Effective Language Models for Recommendation.arXiv preprint arXiv:2408.08821(2024)

    2024

  31. [40]

    Xubin Ren, Wei Wei, Lianghao Xia, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. 2024. Representation learning with large language models for recommendation. InWWW. 3464–3475

    2024

  32. [41]

    Leheng Sheng, An Zhang, Yi Zhang, Yuxin Chen, Xiang Wang, and Tat-Seng Chua

  33. [42]

    Jacob Mitchell Springer, Suhas Kotha, Daniel Fried, Graham Neubig, and Aditi Raghunathan. 2024. Repetition improves language model embeddings.arXiv preprint arXiv:2402.15449(2024)

    2024 arXiv

  34. [43]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang

  35. [44]

    Language Representations Can be What Recommenders Need: Findings and Potentials.arXiv preprint arXiv:2407.05441(2024)

    2024 arXiv

  36. [45]

    Zhulin Tao, Xiaohao Liu, Yewei Xia, Xiang Wang, Lifang Yang, Xianglin Huang, and Tat-Seng Chua. 2022. Self-supervised learning for multimedia recommenda- tion.IEEE Transactions on Multimedia25 (2022), 5107–5116

    2022

  37. [46]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models.arXiv preprint arXiv:2302.13971(2023)

    2023 arXiv

  38. [47]

    Mengting Wan and Julian McAuley. 2018. Item recommendation on monotonic behavior chains. InRecSys. 86–94

    2018

  39. [48]

    Jiaxi Tang and Ke Wang. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. InWSDM. 565–573

    2018

  40. [49]

    Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training.arXiv preprint arXiv:2212.03533(2022)

    2022 arXiv

  41. [50]

    Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2023. Improving text embeddings with large language models.arXiv preprint arXiv:2401.00368(2023)

    2023 arXiv

  42. [51]

    Xiang Wang, Xiangnan He, Meng Wang, Fuli Feng, and Tat-Seng Chua. 2019. Neural graph collaborative filtering. InSIGIR. 165–174

    2019

  43. [52]

    Mengting Wan, Rishabh Misra, Ndapa Nakashole, and Julian McAuley. 2019. Fine-grained spoiler detection from large-scale review corpora.arXiv preprint arXiv:1905.13416(2019)

    2019 arXiv

  44. [53]

    Yinwei Wei, Xiang Wang, Qi Li, Liqiang Nie, Yan Li, Xuanping Li, and Tat-Seng Chua. 2021. Contrastive learning for cold-start recommendation. InMM. 5382– 5390

    2021

  45. [54]

    Yinwei Wei, Xiang Wang, Liqiang Nie, Xiangnan He, Richang Hong, and Tat-Seng Chua. 2019. MMGCN: Multi-modal graph convolution network for personalized recommendation of micro-video. InProceedings of the 27th ACM international conference on multimedia. 1437–1445

    2019

  46. [55]

    Jiancan Wu, Xiang Wang, Fuli Feng, Xiangnan He, Liang Chen, Jianxun Lian, and Xing Xie. 2021. Self-supervised graph learning for recommendation. InSIGIR. 726–735

    2021

  47. [56]

    Wei Wei, Xubin Ren, Jiabin Tang, Qinyong Wang, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. 2024. Llmrec: Large language models with graph augmentation for recommendation. InSIGKDD. 806–815. LLM2Rec: Large Language Models Are Powerful Embedding Models for Seq...

    2024

  48. [57]

    Xu Xie, Fei Sun, Zhaoyang Liu, Shiwen Wu, Jinyang Gao, Jiandong Zhang, Bolin Ding, and Bin Cui. 2022. Contrastive learning for sequential recommendation. In ICDE. IEEE, 1259–1273

    2022

  49. [58]

    An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Cheng- peng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, Jianxin Yang, Jin Xu, Jingren Zhou, Jin...

    2024 arXiv

  50. [59]

    Fajie Yuan, Xiangnan He, Alexandros Karatzoglou, and Liguang Zhang. 2020. Parameter-efficient transfer from sequential behaviors for user modeling and recommendation. InSIGIR. 1469–1478

    2020

  51. [60]

    Yunjia Xi, Weiwen Liu, Jianghao Lin, Xiaoling Cai, Hong Zhu, Jieming Zhu, Bo Chen, Ruiming Tang, Weinan Zhang, and Yong Yu. 2024. Towards open-world recommendation with knowledge augmentation from large language models. In RecSys. 12–22

    2024

  52. [61]

    Chao Zhang, Shiwei Wu, Haoxin Zhang, Tong Xu, Yan Gao, Yao Hu, and En- hong Chen. 2024. NoteLLM: A Retrievable Large Language Model for Note Recommendation. InWWW. 170–179

    2024

  53. [62]

    Junjie Zhang, Ruobing Xie, Yupeng Hou, Xin Zhao, Leyu Lin, and Ji-Rong Wen

  54. [63]

    Xinyu Zhang, Linmei Hu, Luhao Zhang, Dandan Song, Heyan Huang, and Liqiang Nie. 2024. Laser: Parameter-Efficient LLM Bi-Tuning for Sequential Recommen- dation with Collaborative Information.arXiv preprint arXiv:2409.01605(2024)

    2024 arXiv

  55. [64]

    Zheng Yuan, Fajie Yuan, Yu Song, Youhua Li, Junchen Fu, Fei Yang, Yunzhu Pan, and Yongxin Ni. 2023. Where to go next for recommender systems? id-vs. modality-based recommender models revisited. InSIGIR. 2639–2649

    2023

  56. [65]

    Feng Zhu, Yan Wang, Chaochao Chen, Jun Zhou, Longfei Li, and Guanfeng Liu

  57. [67]

    Recommendation as instruction following: A large language model em- powered recommendation approach.TOIS(2023)

    2023

  58. [69]

    Kun Zhou, Hui Wang, Wayne Xin Zhao, Yutao Zhu, Sirui Wang, Fuzheng Zhang, Zhongyuan Wang, and Ji-Rong Wen. 2020. S3-rec: Self-supervised learning for sequential recommendation with mutual information maximization. InCIKM. 1893–1902

    2020

  59. [2019]

    BERT4Rec: Sequential recommendation with bidirectional encoder repre- sentations from transformer. InCIKM. 1441–1450

  60. [2021]

    KDD ’25, August 3–7, 2025, Toronto, ON, Canada Yingzhi He, Xiaohao Liu, An Zhang, Yunshan Ma, and Tat-Seng Chua A Implemental Details A.1 Training Pseudo-code

    Cross-domain recommendation: challenges, progress, and prospects.arXiv preprint arXiv:2103.01696(2021). KDD ’25, August 3–7, 2025, Toronto, ON, Canada Yingzhi He, Xiaohao Liu, An Zhang, Yunshan Ma, and Tat-Seng Chua A Implemental Details A.1 Training Pseudo-code. Here we provi...

    2021 arXiv

  61. [2023]

    InRecSys

    Tallrec: An effective and efficient tuning framework to align large language model with recommendation. InRecSys. 1007–1014

  62. [2024]

    Bridging language and items for retrieval and recommendation.arXiv preprint arXiv:2403.03952(2024)

    2024 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.